From 2e56e28c5dadfed888e859ebbaff09e504351132 Mon Sep 17 00:00:00 2001
From: wangdi <wangdi>
Date: Fri, 20 Jul 2007 17:49:42 +0000
Subject: [PATCH] Branch: HEAD 1) add client stats collection in lstat. 2) add
 README

---
 lustre-iokit/stats-collect/README    | 75 ++++++++++++++++++++++++++++++++++++
 lustre-iokit/stats-collect/lstats.sh | 67 ++++++++++++++++++++++++++++++++
 2 files changed, 142 insertions(+)
 create mode 100644 lustre-iokit/stats-collect/README

diff --git a/lustre-iokit/stats-collect/README b/lustre-iokit/stats-collect/README
new file mode 100644
index 0000000..28e4d12
--- /dev/null
+++ b/lustre-iokit/stats-collect/README
@@ -0,0 +1,75 @@
+Overview
+--------
+These script will be used to collect profile info of lustre client and server.
+It will be run on a single(control) node, and collect all the profile info and 
+create a tarball on the control node. 
+
+lstat.sh : The stat script for single node, which will be run on each profile 
+	   node.
+gather_stats_everywhere.sh : collect stats script.
+config.sh : the config for gather_stats_everywhere.sh.
+
+Requirements
+-------
+1) Lustre is installed and setup on your profiling cluster.
+2) ssh/scp to these node names works without requiring a password.
+
+Configuration
+------
+Configuration is very simple for this script, all of the profiling config VARs are
+in config.sh
+
+XXXX_INTERVAL: the profiling interval
+where value of interval means:
+   0 - gather stats at start and stop only
+   N - gather stats every N seconds
+if some XXX_INTERVAL isn't specified, related stats won't be collected
+XXXX can be: VMSTAT, SERVICE, BRW, SDIO, MBALLOC, IO, JBD, CLIENT 
+
+As for ior-collect-stat.sh, you can modify the various IOR and MPI 
+parameters inside ior-collect-stat.sh 
+
+Running
+--------
+1) The gather_stats_everywhere.sh will be run in three mode
+   
+   a)sh gather_stats_everywhere.sh config.sh start 
+     It will start collect stats on each node provided in config.sh
+   
+   b)sh gather_stats_everywhere.sh config.sh stop <log_name>
+     It will stop collect stats on each node. If <log_name> is provided,
+     it will create a profile tarball /tmp/<log_name>.tar.gz.
+   
+   c)sh gather_stats_everywhere.sh config.sh analyse log_tarball.tar.gz csv
+     It will analyse the log_tarball and create a csv tarball for this
+     profiling tarball. 
+
+2) The ior-collect-stat.sh will be run as
+        sh ior-collect-stat.sh start <profile> 
+   It will create a ior result csv file. If <profile> is provided, 
+   the detail profile info tarball will be created under /tmp.
+
+Example
+-------
+When you want collect your profile info, you should
+   1)sh gather_stats_everywhere.sh config.sh start 
+	 #start the collect profile daemon on each node.
+
+   2)run your test.
+
+   3)sh gather_stats_everywhere.sh config.sh stop log_tarball
+     #stop the collect profile daemon on each node, cleanup
+      the tmp file and create a profiling tarball.
+
+   4)sh gather_stats_everywhere.sh config.sh analyse log_tarball.tar.gz csv
+     #create a csv file according to the profile.
+
+TBD
+------
+Add liblustre profiling support and add more options for analyse.  
+
+
+
+   
+
+
diff --git a/lustre-iokit/stats-collect/lstats.sh b/lustre-iokit/stats-collect/lstats.sh
index b0a04bd..217144d 100755
--- a/lustre-iokit/stats-collect/lstats.sh
+++ b/lustre-iokit/stats-collect/lstats.sh
@@ -272,6 +272,72 @@ function service_start()
 }
 
 #
+# client_stats collector
+#
+# CLIENT_INTERVAL:
+# - 0 - collect at start and stop only
+# - N - collect each N seconds
+#
+function client_collector()
+{
+	local file=$1
+	local target=$2
+	local srv=$3
+
+	echo "client stats for ${target}/${srv} " `date`
+
+	# clear old stats
+	echo 0 >$file
+
+	if let "CLIENT_INTERVAL==0"; then
+		grep -v "^[^ ]*[^0-9]*0 samples" $file
+		idle_collector
+		grep -v "^[^ ]*[^0-9]*0 samples" $file
+	elif let "CLIENT_INTERVAL>0"; then
+		while [ "$stop_collector" != "1" ]; do
+			grep -v "^[^ ]*[^0-9]*0 samples" $file
+			sleep $CLIENT_INTERVAL
+		done
+	else
+		echo "Invalid CLIENT_INTERVAL=$CLIENT_INTERVAL"
+		idle_collector
+	fi
+}
+
+function client_start()
+{
+	if [ "$CLIENT_INTERVAL" == "" ]; then
+		return;
+	fi
+
+	# find all osc 
+	for i in /proc/fs/lustre/osc/* ; do
+		target=`basename $i`
+		if [ "$target" == "num_refs" ]; then
+			continue;
+		fi
+		for j in ${i}/*; do
+			stats=`basename $j`
+			if [ "$stats" == "stats" -o "$stats" == "rpc_stats" ]; then
+				run_collector "osc-${stats}" client_collector \
+					${j} $target $stats &
+			fi
+		done
+	done
+	# find all llite stats
+	for i in /proc/fs/lustre/llite/* ; do
+		target=`basename $i`
+		for j in ${i}/*; do
+			stats=`basename $j`
+			if [ "$stats" == "stats" -o "$stats" == "vfs_ops_stats" ]; then
+				run_collector "llite-${stats}" client_collector \
+					${j} $target ${stats} &
+			fi
+		done
+	done
+}
+
+#
 # sdio_stats collector
 #
 # SDIO_INVERVAL:
@@ -529,6 +595,7 @@ function ls_start()
 	mballoc_start
 	io_start
 	jbd_start
+	client_start
 }
 
 #
-- 
1.8.3.1